LinkedIn's AI Slop Button: Using AI Without Getting Flagged

In this article

1. What the button does to a post
2. What LinkedIn counts as slop, and what it allows
3. A drafting workflow that starts from real work
4. 1. Write the note before the post
5. 2. Give the AI the note and forbid invention
6. 3. Strip the stock phrasing
7. 4. Review by hand, out loud
8. When the analytics flag shows up
9. How HireSteward handles this
10. Self-audit checklist before you publish

LinkedIn now lets any member mark a post as "Seems like AI slop" from the three-dot menu, and those reports train the models that decide how far a post travels. Posts those models classify as slop are less likely to be shown beyond the author's direct connections. That hurts most if you draft with AI and publish what comes back, and the fix starts with the material you hand the AI before it writes anything.

What the button does to a post

The option sits in the three-dot menu at the top right of any post or comment, as Social Media Today describes it. TechCrunch dates the announcement to July 30, 2026. CNN's coverage, as summarized by Slashdot, says the button rolled out in August. The exact launch day matters less than what a report does.

A report does two things:

• It hides the post from the person who reported it. That member sees "Thanks for letting us know. Your feedback helps improve the feed," according to ACS Information Age.
• It becomes training data for LinkedIn's slop detection. LinkedIn built a team made up largely of former journalists and social media experts to teach its model what authentic, useful posts look like. A post the model flags as low-quality AI content is less likely to be distributed beyond the author's direct connections.

Chief Product Officer Hari Srinivasan said the button had been used more than a million times. He also said members were seeing 40% fewer views on what LinkedIn classifies as AI slop than a few weeks earlier (Engadget). The Register read that as people who copy-paste AI posts getting about 40% fewer views. That reading goes further than the data supports. As Superpower Daily points out, the figure covers slop views overall, reflects both reports and detection changes, and does not isolate the button's effect. No single report decides how your post is distributed, and LinkedIn says it has built safeguards so individual feedback can't be used to unfairly target members.

LinkedIn has also left a lot unsaid. Per Depost's review of the record, it has not published how detection works, how accurate it is, or how many flags it takes before reach changes. It has not described any account-level penalty for AI-written posts. Treat the flag as a per-post distribution problem. Nothing reported so far deletes posts or restricts accounts.

What LinkedIn counts as slop, and what it allows

The working definition came from VP of Product Laura Lorenzetti in May 2026. According to MagicPost's summary, she described AI slop as content that "may sound polished on the surface but lacks any real unique perspective or substance." She also wrote that "It's ok to use AI to help you write, but your posts and comments need to represent your voice and your perspectives." Engadget reported that the change would reduce reach for engagement bait, recycled thought leadership, generic content, and posts with obvious AI construction, such as the formula that denies one claim and then swaps in another.

Two later moves show where LinkedIn draws the line. It removed its built-in "enhance" writer and replaced it with a feature that "proofreads your words, but does not change your voice". And Editor-in-Chief Dan Roth told CNN, "I know you think you're doing the right thing with your perfectly bulleted list and your hook at the top... But what works is being human."

So the answer to whether LinkedIn penalizes AI content is narrower than either camp claims. It allows AI-assisted writing and limits reach for content it classifies as polished but empty.

If you've noticed a reach drop on AI-drafted content since late July, phrasing is probably the smaller cause. MagicPost compared each post with the same author's other posts. Posts leaning on that deny-then-swap contrast got about 4.9% lower reach in its English data, roughly the same as in 2025 (MagicPost). It also says that since late July the posts losing the most ground are the ones with nothing concrete in them, however they're phrased. This is vendor data, so weigh it accordingly.

ZoomSphere argues that LinkedIn can't detect AI writing at all and only sees engagement. That contradicts LinkedIn's own statements about training classifiers. The practical advice holds under either model, because a post with no specific number, decision or opinion gets flagged by the classifier or ignored by readers. Scrubbing phrasing is still worth doing, but it is the smaller lever.

A drafting workflow that starts from real work

This order keeps AI useful without letting it supply the substance.

1. Write the note before the post

Keep a running file of work notes. Write each one when something happens, three to five lines in your own words. Useful note types:

• A metric with a before and after, and what changed it
• An incident: what broke, how you found it, what you changed
• A decision, the option you rejected, and why
• A mistake and what it cost

An illustrative note might read: "Retried webhook charged a customer twice. Keyed the charge on the order id. Cost: one extra lookup per payment." That is enough for a post because it holds a mechanism, a fix and a trade-off. If there is no note, don't write the post.

2. Give the AI the note and forbid invention

Paste the note and tell the model to use only facts from it. Ask it to leave a placeholder such as [number?] wherever a detail is missing instead of filling the gap. Then check every number, name and event in the draft against the note.

Invented anecdotes produce the polished surface without perspective that LinkedIn's definition describes. They also fail with readers who know you, who will spot a client story that never happened.

3. Strip the stock phrasing

Edit out the shapes readers now associate with AI-generated LinkedIn posts:

• Contrasts that set up one claim only to knock it down and swap in another
• Hooks built from a template instead of from the note
• Every paragraph broken into bullets
• A closing line that turns the point into a slogan
• Engagement bait at the end, such as asking readers to comment a word

Per the MagicPost data above, this step alone recovers a few percent at most. If the post has nothing in it, rewording won't save it. Go back to step 1.

4. Review by hand, out loud

Roth said some creators draft with tools such as Claude or Gemini and then ask why the posts performed poorly. His creator team, which he calls "content therapists," asks two questions (ProPakistani). Did you enjoy reading it, and does it sound like you?

Read the draft aloud and ask both. If either answer is no, rewrite the failing parts yourself. This is the step that lets you write LinkedIn posts in your own voice with AI help. After it, AI can proofread for typos and tighten a sentence, the same scope LinkedIn gave its own tool. It should not rewrite paragraphs.

When the analytics flag shows up

LinkedIn is adding a private notice in post analytics that reads "some members told us this post seems like AI" (Engadget). At launch Srinivasan described it as a test (Social Media Today), so you may not see it yet.

When it appears:

1. Open the flagged post next to two or three of your posts that did well.
2. List what the good posts had that this one lacks: a number, a decision, an opinion, or a detail only you could know.
3. Check whether you published this draft without the hand review in step 4.
4. Fix the process for the next post. Rewording the flagged post won't add what was missing.

If you run posting for a team or clients, keep two cautions in mind. Anyone can flag a post, and AI detectors give probabilistic estimates with high false-positive rates that disproportionately misclassify writing by non-native English speakers (Cybernews). Readers are also primed to suspect AI. A Pangram study of more than a million opted-in posts found that 41% of LinkedIn posts over 250 words were fully AI-generated, the highest of the platforms studied (The Register).

Spread posting across the real people whose names are on the posts. Don't push copy-pasted AI drafts at scale, and don't treat a clean detector score as protection.

How HireSteward handles this

HireSteward follows the workflow above. You keep a bank of notes about real work (projects, mistakes, decisions), and the writer builds posts only from those notes, never inventing anecdotes. Voice profiles learn your tone from your past posts. Drafts are checked for AI-sounding phrasing before they reach you, and every post can be reviewed and edited before it is published to LinkedIn or X.

It doesn't replace step 4. You still read each post before it goes out. The Free plan allows 3 posts per month and needs no credit card, so you can start free and test it against your own notes.

Self-audit checklist before you publish

• The post comes from a real work note you can point to.
• It contains a detail only you could know.
• Every number in it matches the note.
• It states a decision or opinion with a reason.
• No anecdote, customer or result in it was invented.
• It has no setup-and-swap contrasts, template hooks or slogan endings.
• You read it aloud and it sounds like you talking.
• AI proofread it after your edit and did not rewrite it.

Frequently asked questions

Does LinkedIn penalize AI content?

LinkedIn says using AI to help you write is allowed as long as the post represents your voice and perspective. Content its models classify as low-quality AI slop is less likely to be shown beyond your direct connections. LinkedIn has not described any account-level penalty for AI-written posts.

What happens when someone clicks Seems like AI slop on my post?

The post is hidden from that member's feed, and the report is used to train LinkedIn's slop detection models. LinkedIn says no single report decides how a post is distributed and that it has safeguards against individual feedback being used to unfairly target members.

How will I know if my post was reported as AI slop?

LinkedIn is adding a private notice in post analytics that says some members told us this post seems like AI. It was described as a test at launch, so not every author may see it yet. Only you see it, not your audience.

Will removing AI phrases fix my reach?

It helps a little, but vendor data suggests formula phrasing costs only a few percent of reach. The posts losing the most ground are the ones with nothing concrete in them. Adding a real number, decision or detail from your own work matters more than rewording.

Can I still use AI to write LinkedIn posts?

Yes. The safer pattern is to start from your own notes, have AI structure or tighten them without inventing anything, and then review the draft by hand. LinkedIn's own tool now proofreads without changing your voice, which is a useful model for how far to let AI go.

Sources

1. LinkedIn offers the option to report AI slop (Social Media Today)
2. LinkedIn adds a button to report AI-generated 'slop' (TechCrunch)
3. LinkedIn Fights AI Slop, Now 40% of Its Long-Form Posts (Slashdot, summarizing CNN)
4. LinkedIn adds 'Seems like AI slop' button (ACS Information Age)
5. Even LinkedIn is getting tired of your AI-written posts (CNN)
6. LinkedIn says its AI slop button is working (Engadget)
7. Users mash LinkedIn's AI slop button 1M+ times in 3 weeks (The Register)
8. LinkedIn details AI-slop crackdown after a million users flag posts (Superpower Daily)
9. Does LinkedIn Penalize AI Content? The 2026 Record (Depost)
10. Does LinkedIn Penalize AI Content? What the Data Says (2026) (MagicPost)
11. LinkedIn doesn't want your AI slop anymore (Engadget)
12. LinkedIn Moves To Reduce AI-generated Content (Dataconomy)
13. "It's Not X, It's Y": LinkedIn's Most-Flagged AI Formula (MagicPost)
14. LinkedIn Algorithm 2026: How It Ranks Posts and Why Your Reach Dropped (ZoomSphere)
15. LinkedIn Has the Highest Amount of AI Slop Among Major Platforms (ProPakistani)
16. Those polished LinkedIn posts? Study says chances are they're AI-generated (Cybernews)
17. AI slop writing has taken over the internet, particularly LinkedIn and X (The Register)